feature: Paragraph.list_format

[tomron]

I have a the simplest docx file you can think of - text line with numbering (1. abc) and text line with bullet (* abc) the output for reading the file content with -
print '\n\n'.join([paragraph.text for paragraph in document.paragraphs])

is -
ABC

ABC

I.e it ignores the bullet and the numbering, what is the best way to get the bullets \ numbering? (I use your package via textract although I checked it directly on your package, so I look for a way to do it without diving into the formatting)

[scanny]

In Word, list markers, including bullets and numbers, are aspects of the list, not the paragraphs in the list.

If you wanted to add a character list marker to a paragraph's text when the paragraph was in a list, the simple approach would be to test for a list style:

if paragraph.style = 'ListBullet':
    text = '* %s' % paragraph.text

It gets tricky though, because there are several ways a paragraph can end up appearing in a list. For that you'd need to identify those cases and trace through to the numbering definition to see what bullet type was used in each case.

[tomron]

Thanks for the fast answer. Basic listing - just finding out if something is in a list is good enough for me at this point. However, I tried it on this simple document (created it once using libreoffce and saving it as docx and another time using google document and downloading it as docx) and in both cases style was "Normal". I don't know what I do wrong

[scanny]

What you're seeing I expect is a numbering definition being applied directly to the paragraph rather than via a style. python-docx doesn't have to capabilities to parse out those bits at the moment. The higher priority initially has been to enable document generation rather than analysis and modification of existing documents, although many features are useful for both.

There's a bit more on what's involved here in this SO answer:
http://stackoverflow.com/questions/23446268/python-docx-how-to-restart-list-lettering/23464442#23464442

You can add a feature request for it, or we can just turn this issue into a feature request. I think the right name would be 'feature: detect directly-applied numbering'.